Fantasy Football Advice

With the NFL season growing near, many are gearing up for their fantasy football drafts. Being both and economics and sports nerd, I decided to try and figure out what best determines how many fantasy points a player will earn. In order to do so, I have run three separate regressions (one for QBs, one for RBs, and one for WRs and TEs).

The regressions were as follows:

For QBs:

Pts = B0 + B1(games) + B2(Experience) + B3(BCS dummy) + B4(YdsGmTD) + B5(WinPct) + B6(OppWinPct) + B7(TDPct) + e

For RBs:

Pts = B0 + B1(Games) + B2(Experience) + B3(TDpct) + B4(YdsGmTD) + B5(RecYds) + B6(RecTD) + B7(WinPct) + B8(OppWinPct) + e

For WR/TEs:

Pts = B0 + B1(Games) + B2(Experience) + B3 (BCS dummy) + B4(YdsGmTD) + B5(TDpct) + B6(WinPct) + B7(OppWinPct) + e

Where TDpct is the amount of touchdowns per passing/rushing attempt or catch, YdsGmTD is yards per game plus touchdowns, BCS dummy is a dummy variable for whether or not the player was a member of a college football team affiliated with the Bowl Championship Series. (The regressions were also ran without the “Games” variable.)

The results explained the following:

1. The coefficient on TD percentage was the largest of all significant variables.
2. College background is not significant.
3. Yards per game + TD was significant in each regression.
4. The deviation in points amongst WR/TEs was smaller than both RBs and QBs. This means that one should not waste their first pick on a WR because there is less to gain relative to other positions.

The graphs below correspond to QB, RB, and WR/TE respectively:

These regressions are open to criticism. However, the results are very intuitive and thus I am inclined to believe that they may be helpful. So, if you are drafting your team soon, keep TD Pct in mind.

Good luck!